4 Lexical Accents and Head Dominance in Fusional Languages Greek and Russian 4.1. Introduction What primarily identifies a lexical accent system is the notion of competition between lexical accents for primary stress. We have seen in Chapter 3 that for the greatest part of the vocabulary, prosodic structure is determined by the inherent properties of morphemes. Prosodic principles are only employed to restrict the freedom of marking. Still, given the fact that in the languages examined here one primary stress is allotted to each morphological word,1 the question is how accentuation is pursued when more than one marked morpheme is present in a word. This question is undertaken in the present chapter, which focuses on the morphological aspect of lexical accent systems. More specifically, the proposal is that stress depends on morphological structure and especially, the hierarchical relations that hold between the elements of the word. Given the fact that morphological structure plays a cardinal role for stress assignment, we expect languages that employ different morphological mechanisms to build up their words to diverge in the way they pursue accentuation. In this chapter, I am primarily concerned with lexical accent systems of fusional morphology and in particular with Greek and Russian. In Chapter 5, I focus on lexical accent systems of polysynthetic morphology. In fusional languages, roots combine with several affixes to form words. As a consequence, words minimally consist of two morphemes, a root and an affix. Morphological complexity is an expected property in lexical accent systems. It is the rich morphology that brings to light the inherent accentual properties of 1 This is triggered by a high ranked constraint which, in general, states that each prosodic word has one prominent constituent. This constraint is low ranking in pitch-accent systems. For example, in Tahltan all accents in a word can bear a high tone, e.g. k’í7hédeØs-déØl ‘they three or more run’. 172 CHAPTER 4 morphemes and, eventually, the dependence of prosody on morphological structure. 4.1.1. Theoretical explorations in Chapter 4 This section provides a brief introduction to the main ideas advanced in chapter 4. As mentioned in the previous chapter, most morphemes are stored in the lexicon with a lexical accent. Moreover, inherent metrical information supersedes the phonological constraints that together constitute the ‘default accentuation’. However, how is accentuation pursued in more intricate metrical constructions? What happens when two or three marked morphemes are present in the word? Which accent prevails as primary? In Chapter 1, I claimed that head-oriented systems underline the significance of morphological structure by segregating head from non-head constituents and assigning prominence to heads. Generally speaking, there are two varieties of head-based systems; head-stress systems and head-dependent systems. In the former type, a morphological head is obligatorily assigned prominence, whereas in the latter type a head prevails only when it is marked. When the head lacks inherent accentual properties other marked constituents are given a chance to determine stress. Marking and morphological structure, and particularly the notion ‘head of the word’, are vital components for the accentuation of head-dependent systems like Greek and Russian. The specifics of marking were examined in Chapter 3. The focus here is on the morphological component of lexical accent systems with special emphasis on the role of headedness for stress. The central claim in this chapter is that competing accents represent competing morphemes. More specifically, when two accents occur in a word, the accent introduced by the morphological head is prosodically prominent.2 Headedness must be interpreted in a strict fashion, meaning the ability of a morpheme to determine the word’s syntactic category. A derivational suffix that changes the base it is attached to from nominal to adjectival is considered to be a head. In the same spirit, roots are heads in inflected words because they determine the syntactic category of the whole form (Zwicky 1985, Scalise 1988a, among others). Assigning primary stress to the morphological head means that the inherent accentual properties of roots outrank the inherent properties of inflectional suffixes in inflected constructions, but submit to the inherent metrical information of derivational suffixes in derived constructions. 2 This idea has been proposed for Greek stress by Ralli (1988) and Ralli and Touradzidis (1992) (cf. §3.3.2). HEAD DOMINANCE IN FUSIONAL LANGUAGES 173 In short, there is a split in the accentual behavior of marked morphemes; heads are given priority for stress, provided that they bear an accent. This claim is supported by the empirical facts of Greek and Russian inflectional and derivational morphology.3 I will illustrate the above with some examples. The Greek root /sta(fi'-/ and the genitive plural inflectional suffix /-(on/ are accented. When these morphemes join to form a word, stress falls on the root, stafí'on. This implies that the accent of the inflectional ending yields to the accent of the root. On the other hand, an accented derivational suffix such as /-(ini/ attracts stress from the root stafi'íni ‘raisin pulp’ simply because in the new formative the suffix, and not the root, is the head. If morphological structure is important for prosody, the main question is what principle allows the interface between these components of grammar. I claim that the prosody-morphology interface centers around the principle of compositionality. This principle is borrowed from formal semantics (Montague 1974) and, intuitively, entails that the interface between two levels is established through one and the same structure. For instance, each time a syntactic rule applies to combine two lexical items, the semantic interpretation of the derived expression is determined by the interpretation of the two expressions combined. Similarly, when a morphological rule applies to combine two morphemes, the phonological interpretation of the derived expression is determined by the phonological interpretations of its parts. In interface systems, compositionality or rather, prosodic compositionality simply implies that prosody can have access to morphological structure because the two components of grammar are built in a parallel fashion. It allows prosodic structure to interact with morphological structure and, more importantly, become sensitive to the morphological rules that apply to form various morphological formations (i.e. inflected or derived formations). For instance, because of prosodic compositionality prosody can become sensitive to the morphological rules that build up a head-dependent relation between a root and an inflectional suffix. In lexical accent systems in particular, the prosody- morphology interface is articulated in terms of a theory of head dominance, which states that the accent of the morphological head of the word prevails over other accents. Head dominance enriches Universal Grammar with the family of head constraints which are part of a broader family of interface constraints. These constraints allow a direct relation between prosodic elements and morphological constituents such as, for example, lexical accents and morphological heads. Two 3 I take for granted that morphological constraints of affixation are high ranked in the languages examined in this study. 174 CHAPTER 4 types of head constraints are important in this study: HEADFAITH and HEADSTRESS. Both constraints have been introduced in previous chapters. The former constraint states that a lexical accent sponsored by a (morphological) head should have a correspondent in the output and vice versa, a lexical accent hosted by a (morphological) head must have a correspondent in the input. The latter constraint simply states that a (morphological) head must be stressed. Later in this chapter, I show that head dominance is expressed by means of a ‘positional faithfulness ranking’ in which the more specific faithfulness constraint, HEADFAITH, dominates general faithfulness, FAITH: HEADFAITH >> FAITH. Theoretically, the function that executes the prosody-morphology mapping has an infinite number of interpretations. Greek and Russian choose to phonetically interpret it as stress. Japanese and Hua, on the other hand, interpret prominence as a tonal contour, whereas Turkana realizes the mapping between phonology and morphology by means of harmony. A language may also choose to interpret this function as prominence of the non-head element of the word. However, to my knowledge there are no accentual systems that give systematic priority to non-heads. In this chapter, based on Dresher and Van der Hulst’s (1997) theory of headedness, I make the stronger claim that such systems do not exist. Prosodic compositionality as introduced above, predicts that different morphological structures will have a different impact on stress. This prediction is indeed borne out here as well as in Chapter 5. For instance, in derived words the (marked) derivational suffix prevails over the root and the inflectional suffix because it is the head. In incorporated constructions, on the other hand, the root is the head and the suffix is the complement that incorporates to the root/head. According to what has been proposed so far, in incorporated constructions an accented root will be prosodically dominant. To summarize, prosodic compositionality is the principle that permits the interface